The Fifth Circuit Says TCPA Written Consent Was Never Required. Here’s What That Means for AI Voice Platforms.

The Fifth Circuit Court of Appeals just announced that the TCPA does not require prior express written consent for prerecorded calls. The ruling in Bradford v. Sovereign Pest Control contradicts what the FCC’s regulations have required for over a decade and is probably the biggest development in TCPA consent law since the 11th Circuit struck down the one-to-one consent rule.

Every TCPA commentator will cover what the ruling says. This article focuses on a different question: what does Bradford mean for companies building AI voice technology, and does it actually reduce the platform liability exposure that has been growing since the Lowrey v. OpenAI lawsuit?

The short answer: less than you might hope.

What the Court Actually Said

The facts are straightforward. Mr. Bradford had a pest control contract with Sovereign Pest Control. He gave Sovereign his phone number in case they “needed to get in contact” with him. Sovereign made prerecorded calls to schedule renewal inspections. Bradford renewed four times, then sued alleging TCPA violations.

The Fifth Circuit sided with Sovereign, but the reasoning is what matters. The TCPA statute requires “prior express consent of the called party” for prerecorded calls. The FCC’s regulations added a higher standard for telemarketing calls: “prior express written consent.” The Court found that the FCC exceeded its statutory authority by creating a consent requirement Congress never enacted.

As the Court put it: “Contrary to the FCC’s regulation, Congress permits either written or oral consent for any auto-dialed or pre-recorded call.” The “written” requirement was the FCC’s invention, not Congress’s.

This follows the same post-Loper Bright pattern that produced the one-to-one consent ruling. Courts are reading the statute Congress passed rather than the regulatory structure the FCC built on top of it.

Why AI Voice Builders Are Paying Attention

The FCC’s February 2024 declaratory ruling confirmed that AI-generated voice is “artificial voice” under the TCPA. That means every rule applying to prerecorded calls applies to AI voice agents, including the FCC’s written consent requirement for marketing calls.

If the Fifth Circuit is right that the statute never required written consent, the implication for AI voice platforms is immediate: the consent threshold for AI voice marketing calls may be lower than the industry has been operating under. “Prior express consent” can be oral. It doesn’t require a signed form, a checkbox, or the specific written consent elements the FCC’s regulations demand.

For platforms whose customers are making AI voice calls at scale, that sounds like a significant reduction in compliance burden. Easier consent means more calls are “legal.” More legal calls means less liability exposure for everyone in the chain.

That logic is appealing. It is also incomplete.

The Platform Liability Question: Does Bradford Help?

The Lowrey v. OpenAI lawsuit established a new front in TCPA litigation: platform providers named as defendants not because they made calls, but because they enabled, facilitated, and profited from calls made by their customers. The complaint alleged that Twilio and OpenAI had the technical capability to prevent violations and chose not to implement safeguards.

Under that theory, platform liability depends partly on whether the underlying calls were illegal. If a platform’s customer had valid consent, the calls were not violations, and there is nothing for the platform to be vicariously liable for. So if Bradford lowers the consent threshold from “written” to “express,” doesn’t that mean fewer calls are violations, which means less platform exposure?

In theory, yes. In practice, the reduction is much smaller than it appears, for three reasons.

1. Consent Was Rarely the Only Problem

The TCPA lawsuits that generate the largest settlements and the most platform liability exposure are not typically pure consent cases. They involve Do Not Call violations, where someone on the national DNC registry was called without any consent at all. They involve revocation failures, where a consumer said “stop calling me” and the calls continued. They involve identification and disclosure failures, where the caller did not provide the required disclosures at the beginning of the call.

Bradford addresses the distinction between written and oral consent. It does not help with calls made to consumers who never consented in any form. It does not help with calls made after consent was revoked. And it does not eliminate the disclosure requirements that apply to every AI voice call regardless of consent type.

American Income Life settled for $14 million on DNC violations, not consent deficiencies. The Lowrey complaint alleged calls to consumers who never consented to anything. For the type of high-volume, high-exposure calling that drives platform liability theories, Bradford addresses a secondary issue, not the primary one.

2. The “We’re Just the Platform” Defense Doesn’t Depend on Consent Type

Platform liability under the TCPA is built on the concept of “causing” a call to be initiated. The question courts are asking is whether a platform that provides the infrastructure, knows it is being used for telemarketing, has the technical ability to prevent violations, and chooses not to implement safeguards has “caused” the illegal calls. Said another way, does the platform go beyond “merely providing the platform” and move into “setting up and providing intimate support for their customers’ campaigns and strategies”?

That analysis does not hinge on whether the consent was written or oral. It hinges on whether the platform was so involved in the calling process that the platform could be considered to have initiated the calls. A platform that enables a customer to make 10,000 AI voice marketing calls without any consent verification infrastructure has the same liability exposure regardless of whether the applicable standard is written consent or express consent.

The vicarious liability analysis from Dickson v. Direct Energy and Klassen v. SolidQuote reinforces this point. In Klassen, a company was held liable for ratification despite being four entities removed from the caller, because its agent proceeded with a sale after learning the call was outbound. The type of consent required was irrelevant to the ratification analysis. What mattered was the real-time decision to continue.

3. The Regulatory Walls Are Getting Higher, Not Lower

Even if the federal TCPA consent floor drops in additional circuits, AI voice platforms face a layered compliance environment that Bradford does not touch.

The FCC’s pending NPRM proposes mandatory AI-specific consent disclosures and in-call disclosure that AI technology is being used. These are new requirements layered on top of existing consent obligations, not dependent on whether that consent is written or oral.

State-level AI regulations are advancing independently. Colorado’s AI Act, effective July 2026, classifies AI voice agents used for insurance and other consequential decisions as “high-risk” regardless of federal TCPA consent standards. State mini-TCPA statutes in Florida and other states impose calling restrictions and statutory damages that stack on top of federal exposure.

And the FTC has made AI voice cloning an enforcement priority, operating under consumer protection authority that is entirely separate from the TCPA consent framework.

A platform builder reading Bradford and concluding “our consent obligations just got easier” is seeing one tree while missing the forest. The consent standard for one type of TCPA claim may have shifted in one circuit. The overall regulatory pressure on AI voice platforms has not decreased.

What Bradford Actually Does Mean for AI Voice Platforms

The ruling is not meaningless for the AI voice ecosystem. It matters in several specific ways.

It provides a new litigation defense in the Fifth Circuit. For AI voice platforms with customers operating in Texas, Mississippi, and Louisiana, the written consent requirement may no longer apply to marketing calls. This is a genuine narrowing of TCPA exposure in that jurisdiction. If the 11th Circuit follows suit, as is likely given their one-to-one consent reasoning, the geographic scope expands.

It changes the consent chain analysis at the margins. The consent chain-of-custody problem, where consent degrades as leads pass through multiple intermediaries, is a major liability vector for AI voice platforms. If the applicable standard is “express consent” rather than “express written consent,” the evidentiary burden shifts. A platform’s customer who can demonstrate the consumer gave verbal consent (recorded call, for example) may have a viable defense even without a signed form. This marginally reduces the number of calls that are clearly non-compliant.

It accelerates the regulatory fragmentation AI voice platforms must navigate. Add “written consent vs. express consent” to the existing circuit splits on “texts as calls” and “ATDS definitions.” Platforms operating nationally now face more jurisdictional variability, not less. This makes compliance architecture more important, not less important, because you cannot build a single national consent standard when different circuits apply different rules.

What AI Voice Platforms Should Do Now

Do not lower your consent standard. Written consent costs you almost nothing to maintain. Oral consent is harder to document, harder to produce in litigation, and not accepted outside the Fifth Circuit. The asymmetry of the risk calculation is overwhelming. Keep getting written consent.

Ensure your consent verification infrastructure works under either standard. If your platform verifies consent before enabling AI outreach, the verification system should be agnostic to whether that consent is written or oral. A well-designed system captures the consent record, timestamps it, associates it with the contact, and produces it on demand. This protects you regardless of which consent standard a court ultimately applies.

Focus your compliance investment on the exposures Bradford does not address. DNC scrubbing. Revocation handling. AI disclosure requirements. In-call identification. These are the compliance failures that drive the largest TCPA settlements and the most aggressive platform liability theories. They are completely unaffected by the consent type distinction Bradford addresses.

Track the circuit split. If the 11th Circuit reaches the same conclusion, the combined geographic scope of the “oral consent is sufficient” position becomes significant. If additional circuits follow, the pressure on the Supreme Court or Congress to resolve the split increases. This is a development to monitor, not a development to act on.

Prepare for the next regulatory layer, not the last one. The FCC’s pending NPRM on AI-specific consent disclosures, state AI disclosure laws, and the evolving platform liability doctrine under Lowrey are all moving forward regardless of Bradford. The platforms that will navigate this environment successfully are the ones building compliance architecture that adapts to changing requirements, not the ones optimizing for the most favorable reading of a single circuit’s opinion.

The Bottom Line

Bradford v. Sovereign Pest Control is a significant TCPA development. For AI voice platforms, it is a meaningful data point in the evolving consent landscape. It is not a license to relax compliance standards.

The platform liability theories being tested in Lowrey v. OpenAI do not depend on whether consent was written or oral. They depend on whether platforms had knowledge, capability, and incentive to prevent violations and chose not to act. That calculus has not changed.

The AI voice platforms that will thrive are the ones treating compliance as infrastructure, not as a reaction to the latest court opinion. Bradford may eventually contribute to a meaningful reduction in consent requirements. Until that happens across multiple circuits, the smart play is to maintain the higher standard while building systems flexible enough to adapt as the law continues to shift.

This should be an interesting year.